Previous studies using retinal ganglion cell neurons as a model system have demonstrated that the phosphate groups on the N-terminal head domain of the 145 kDa middle subunit of neurofilaments (NF-M) in vivo are added by second messenger-dependent protein kinases. In the present study, ser-23 was identified as a specific protein kinase A phosphorylation site on native NF-M subunit and on two synthetic peptides S1 (14RRVPTETRSSF24) and S2 (21RSSFSRVSGSPSSGFRSQSWS41) localized within the N-terminal head domain region. Ser-23 was the major site of phosphorylation for protein kinase A on the 32P-labeled peptides from native NF-M subunit and on both the S1 and S2 peptides. The synthetic peptide S2 also showed low level phosphorylation of ser-28 and ser-32 by protein kinase A. Ser-23, ser-25, ser-28, ser-32 and a threonine residue were phosphorylated by protein kinase C but neither of the site was phosphorylated stoichiometrically. The analyses of the phosphorylated synthetic peptides by LC/MS/MS also showed that protein kinase A phosphorylated only 1 site on peptide S1and ions containing up to 3 phosphates were detected on peptide S2. By comparison, 2 and 4 sites were phosphorylatable by protein kinase C on peptide S1 and S2, respectively. These data showed that ser-23 is a major protein kinase A phosphorylation site on the amino-terminal head domain of NF-M. Since phosphorylation of neurofilaments by protein kinase A in vitro is known to regulate neurofilament assembly/disassembly, phosphorylation of ser- 23 on the amino-terminal head domain may play a critical role in neurofilament dynamics during axonal transport. This hypothesis was supported by a functional assay developed to study how individual protein kinases or protein phosphatases may regulate the assembly/disassembly of various cytoskeletal elements into a supramolecular structure. In this assay, the Triton-insoluble cytoskeleton preparations are treated with 1-3 M urea to release high molecular weight complexes (5 - 20 million daltons) from the cytoskeletal preparation which are then fractionated on a Sephacryl S- 500 column and analyzed by gel electrophoresis and by electron microscopy of rotary shadowed or negatively stained oligomers. For example, the treatment of cytoskeletal preparation with 1.25 M urea resulted in the elution of neurofilament proteins in peaks containing heteropolymers of neurofilament subunits. NF-H eluted in a peak at 8-12 million daltons and NF-M and NF-L peaks eluted at 6-10 million daltons. Similarly, spectrin containing oligomers eluted into two peaks at > 12 million and 4-8 million daltons. Dephosphorylation of the cytoskeletal preparations by calcineurin resulted in more than 100% decrease in the amount of neurofilament proteins dissociated into high molecular weight oligomers. The dephosphorylation-mediated decreased dissociation of neurofilament proteins was partially restored by phosphorylation by protein kinase A but not by protein kinase C both of which phosphorylate only the N-terminal head domain regions of neurofilament proteins. These data support the hypothesis that the supramolecular organization of the axonal cytoskeleton is regulated by site-specific phosphorylation/ dephosphorylation of the constituent cytoskeletal proteins.